The goal of restoring ecological integrity in rivers is frequently accompanied by an assumption that a comparative reference reach can be identified to represent minimally impaired conditions. However, in many regulated rivers, no credible historical, morphological or process-based reference reach exists. Resilient restoration designs should instead be framed around naturalization, using multiple analytical references derived from empirically-calibrated field- and model-based techniques to develop an integrated ecological reference condition. This requires baseline data which are rarely collected despite increasing evidence for systematic deficiencies in restoration practice. We illustrate the utility of baseline data collection in restoration planning for the highly fragmented and regulated lower Merced River, California, USA. The restoration design was developed using various baseline data surveys, monitoring, and modeling within an adaptive management framework. Baseline data assisted in transforming conceptual models of ecosystem function into specific restoration challenges, defining analytical references of the expected relationships among ecological parameters required for restoration, and specifying performance criteria for post-project monitoring and evaluation. In this way the study is an example of process-based morphological restoration designed to prompt recovery of ecosystem processes and resilience. For the Merced River, we illustrate that project-specific baseline data collection is a necessary precursor in developing performance-based restoration designs and addressing scale-related uncertainties, such as whether periodic gravel augmentation will sustain bed recovery and whether piecemeal efforts will improve ecological integrity. Given the numerous impediments to full, historical, restoration in many river systems, it seems apparent that projects of naturalization are a critical step in reducing the deleterious impacts of fragmented rivers worldwide.
